Osteoporosis and fracture as risk factors for self-harm and suicide: a systematic review and meta-analysis

Background Increase in presentations of self-harm to primary care, a risk factor of suicide, has led to a growing interest in identifying at-risk populations. Aim To examine whether osteoporosis or fractures are risk factors for self-harm, suicidal ideation, and suicide. Design and setting This was a systematic review of observational studies in adults (>18 years) that had examined the role of osteoporosis and/or fractures in subsequent self-harm, suicidal ideation, and/or suicide. Method Six databases were searched from inception to July 2019. Additional citation tracking of eligible studies was undertaken in November 2022. Screening, data extraction, and quality assessment of full-text articles were performed independently by at least two authors. Where possible, meta-analysis was run on comparable risk estimates. Results Fifteen studies were included: two examined the outcome of self-harm, three suicidal ideation, and 10 suicide. In approximately half of studies on osteoporosis, the risk of suicidal ideation and suicide remained significant. However, pooling of adjusted odds ratios from three studies indicated no association between osteoporosis and suicide (1.14, 95% confidence interval = 0.88 to 1.49). Nine studies examined the risk of a mixture of fracture types across different outcomes, limiting comparisons. However, all studies examining vertebral fracture (n = 3) reported a significant adjusted negative association for self-harm and suicide. Conclusion Patients with vertebral fractures, a risk potential factor for suicide, may benefit from clinical case finding for mood disorders with personalised primary care management. However, because of the limited number and quality of studies and mixed findings, further examination of these associations is warranted.


INTRODUCTION
Rates of suicide in the UK have risen since 2017, with a 9% increase by 2019. 1 Similarly, there has been an increase in presentations of self-harm to primary care, 2 with the risk of suicide increasing 50-fold in the year after self-harm compared with the general population. 3Such increases have led to a growing interest in identifying at-risk populations.Although a history of mental ill health, notably depressive disorders, is an understood risk factor for self-harm, 4 pain, and more specifically back pain, are also independent risk factors for self-harm 5,6 and suicide, specifically in older adults. 7ecent studies have demonstrated increased risk of self-harm in people with physical health conditions such as fibromyalgia and osteoarthritis. 5,8Although pain is not a symptom of osteoporosis, pain frequently occurs because of subsequent fractures, 9,10 and with one in three women and one in five men aged >50 years experiencing an osteoporotic fracture in their lifetime, osteoporosis has been suggested as a risk factor for self-harm and suicide. 5,11Furthermore, common outcomes of fractures, including osteoporotic fractures, such as isolation and depression are risk factors for suicide. 12revious research examining the risk of self-harm or suicide in older adults with osteoporosis or fractures has yielded conflicting results, 5,[13][14][15] possibly because of a lack of consensus on diagnosis definitions, population variations, and study limitations.The aim of this study was to examine whether osteoporosis and/or fractures are risk factors for self-harm, suicidal ideation, or suicide.

METHOD
A systematic review and meta-analysis was conducted.The protocol was registered on PROSPERO (CRD42021272606) and adhered to PRISMA guidance. 16

Search methods
The population of interest were adults with osteoporosis and/or fractures.The outcomes of interest were risk estimates of self-harm, suicidal ideations (including suicidal behaviours), and suicide.Exclusion criteria were: • studies with a population of <18 years old; e735 British Journal of General Practice, October 2023 • studies that could not be translated; and • systematic reviews, case reports, or case series studies.
Searches were tailored and conducted in six electronic databases: MEDLINE, PsycArticles, AMED, CINAHL Plus, PsycINFO, and Web of Science from inception to July 2019.Search strategies utilised database subject headings and text word searching in title, abstract, or keywords, combining terms for: osteoporosis and fractures; and self-harm, suicidal ideation (including suicidal behaviours), and suicide (Supplementary Information S1).
In addition, the reference lists of included studies and relevant systematic reviews were checked and key studies citation tracked.To account for the delay in publication because of COVID-19, citation tracking was conducted on all included papers in September 2021 and November 2022 to ensure relevant new studies were identified.

Study screening and selection
A two-stage screening of articles against predefined eligibility criteria was implemented: first by titles and abstracts; second by full text.
At each stage, screening was conducted by two reviewers independently (the first, third, and senior author) and articles were excluded by consensus.Arising disagreements were resolved through discussion.

Quality assessment
The Quality in Prognosis Studies (QUIPS) tool was used to assess included articles. 17t examines the risk of bias across six domains: study population, study attrition, prognostic factor measurement, outcome measurement, study confounding, and statistical analysis and reporting.All included articles were assessed for their quality independently by two pairs of assessors (the first and senior author or the second and sixth author).Any disagreement on scoring was initially discussed and arbitrated by a different reviewer if required.

Data extraction
Data were extracted by two authors (the third and fourth authors) on demographic information (age, sex, country of origin, and so on), and size of the study sample, numbers of patients with the condition of interest, study setting (for example, primary care), exposures (for example, osteoporosis or fracture), exposure definition, outcome (for example, self-harm, suicidal ideation, and suicide), and method of risk estimates regarding the association between exposure and outcome.
Findings were tabulated by their exposure category, that is, articles examining either osteoporosis or fracture were compared separately.Within each of these exposure categories, outcomes were subcategorised by either self-harm, suicidal ideation/behaviour, or suicide.Risk estimates, 95% confidence intervals (CIs), and statistical significance were extracted and reported for each study, and where available separately for men and women.
Specifically for fractures, risk estimates were tabulated by location of fracture, where reported.Authors were not contacted if it was not possible to identify certain required information in their publication.

Meta-analysis
Where enough studies (≥3) examining comparable factors were identified, a random-effects meta-analysis was used to pool reported risk of any of the examined exposures along with their 95% CI.Heterogeneity was assessed by I 2 .Analysis was undertaken in Stata (version 14).

Patient and public involvement
Keele Research User Group consists of a diverse group of people with lived experience of osteoporosis, or of caring for people with osteoporosis.
In a series of meetings to discuss research, public contributors talked about the significant psychological and social sequelae of osteoporotic fractures.However, there was no direct public and patient involvement in the conduct or interpretation of this study.

How this fits in
An increased risk of self-harm and suicide has been identified in people with physical health conditions, such as fibromyalgia and osteoporosis.This is the first study, to the authors' knowledge, to synthesise the evidence on osteoporosis and fractures as risk factors for self-harm, suicidal ideation, and suicide (including suicidal behaviours).Although there was no association between osteoporosis and suicide, vertebral fractures appeared to increase the risk of self-harm and suicide.Patients with vertebral fractures therefore may benefit from clinical case finding for mood disorders with personalised primary care management.

Characteristics of osteoporosis studies
Studies were conducted in a range of countries, predominantly Europe 11,13,20 and East Asia. 14,18,19Study settings included primary care, 11 primary and secondary/specialist care, 14,21 and general populations. 13,18,19,20,22Four studies did not report average age and/ or standard deviation (SD), 11,13,19,21 with two only reporting the age ranges used to group participants. 13,19Three studies included participants <45 years of age 11,19,21 and four included participants >80 years of age. 11,13,19,22Osteoporosis was predominantly identified utilising disease codes in a range of health record databases (Supplementary Table S2).In terms of outcomes, suicidal ideation was recorded via patient self-report 14,19,20 and papers reporting numbers of suicide used healthcare records. 11,13,14,21,22teoporosis as a risk factor for suicidal ideation Two of the three studies reported significant crude estimates (crude odds ratios [ORs]), ranging from 1.90 (95% CI = 1.38 to 2.62) to 2.13 (95% CI = 1.34 to 3.39) (Table 2). 18,19cords identified from: Databases  29 Only one of these reported adjusted ORs and they remained significant (for men only 2.07, 95% CI = 1.19 to 3.59, P<0.05). 19he third study did not report a crude ratio; however, they reported a significant adjusted estimate (1.19, 95% CI = 1.04 to 1.36). 20e Kim et al study stratified risk by sex, 19 reporting crude and adjusted risk estimates for suicidal ideation in men and women; however, only risk estimates for males were significant in crude and adjusted ratios (adjusted OR 2.07, 95% CI = 1.19 to 3.59).British Journal of General Practice, October 2023 e738 Osteoporosis as a risk factor for suicide Five of the seven studies relating to osteoporosis as a risk factor for suicide found significant crude results, 11,13,14,21,22 with estimates ranging from a crude OR 1.28 (95% CI = 1.23 to 1.33) 14 to 3.12 (95% CI = 1.56-6.21). 22Four studies also reported adjusted RRs, with two retaining significance (males 1.67, 95% CI = 1.1 to 2.51 and females 1.88, 95% CI = 1.5 to 2.35). 13djusted OR data from a total of 4123 primary and secondary care patients across three studies was pooled and found no association between osteoporosis and subsequent suicide (1.14, 95% CI = 0.88 to 1.49) (Figure 2).

Characteristics of fracture studies
Ten studies, mostly conducted in East Asia 14,15,24,27,28 and Europe 13,20,23 examined fracture as a risk factor.Study settings included primary care, 23 a retirement community, 25 secondary care, 24 across primary and secondary care, 14 and the general population (Table 1). 13,20,26Most studies included a balanced sample of men and women, or a higher percentage of females (as expected in osteoporotic studies).Four studies reported mean ages ranging from 59.8 (SD 14.3) to 75.0 (SD 6.8) years. 14,20,24,26Five studies included participants aged >80 years, 13,24,25,27,28 and one included participants below <35 years (Table 1). 15ractures were defined inconsistently between studies.Five of the 10 studies (examining the outcomes self-harm, 13 suicidal ideation, 15,20 and suicide 14,25 ) reported a grouped fracture risk estimate; however, each utilised a different grouping definition: 'all fractures', 'pathological fractures', 'fractures >50 years', 'other fractures', and 'any fracture in last 5 years' (Table 1 and expanded in Supplementary Table S2).Chang   N/S = outcome was reported as not statistically significant, but no P-value was provided.OR = odds ratio.RR = risk ratio.Sig = outcome was reported as statistically significant, but no P-value was provided.
detail with 11 bodily regions, 14 whereas Tsai et al grouped fractures into just two groups of 'trunk' and 'lower limb' 15 (Supplementary Table S2).Hip and vertebral fractures were the most consistently reported fracture sites, reported in five and four studies, respectively. 13,14,20,23,25,28However, these studies utilised a range of risk estimate types (Table 3).

Fracture as a risk factor for self-harm
Erlangsen et al (2021) 26 investigated 'any fractures in the prior 5 years' as a risk factor for self-harm, whereas Prior et al examined the relationship between vertebral fractures and self-harm categorised by sex. 23Crude and adjusted hazard ratios (HRs) were significant in both men and women.Erlangsen et al (2021) found a significant adjusted incidence ratio (IR) of 1.38 (95% CI = 1.12 to 1.71). 26Prior et al found that men and women with vertebral fractures were almost four and two times more at risk of self-harm than those without such fractures (adjusted HR of 3.90, 95% CI = 1.80 to 8.50 and 1.90, 95% CI =1.10 to 3.20, respectively) (Supplementary Table S3). 23acture as a risk factor for suicidal ideation/behaviour  26 examined the adjusted association between self-reported 'fractures in last 5 years' and suicide but found no significant association (Table 2).Three studies examining the role of vertebral fractures on suicide found significant associations in all, with risk estimates ranging from OR 1.40 (95% CI = 1.07 to 1.82) by Jang et al to RR 2.20 (95% CI = 1.79 to 2.70) by Erlangsen et al (in women). 14,26,28ive studies examined the association between hip/femoral neck fracture and suicide; 14,[24][25][26]28 three of these found a significant crude association. Sigificant associations were retained in two of these after adjustment.13,14 Although several other fracture sites were examined, conflicting or infrequent examination of these means that no consensus on an association with suicide can be drawn (Table 3).Studies examining the risk of self-harm and suicidal ideation in populations with fractures at specific sites were also limited (Supplementary Table S3).

Summary
This systematic review included all identified studies examining osteoporosis and fracture as risk factors for self-harm, suicidal ideation, and suicide.Although data are limited, particularly for self-harm, this study found that for approximately half of osteoporosis and fracture studies there remained significant risks of suicidal ideation or suicide after adjustment.All three studies that examined the role of vertebral fracture as a risk factor for suicide found a significant association, even after adjustment (although data pooling was not  11 Ahmedani et al 21 Chang et al 14 Overall, DL (I 2 = 67.9%,P = 0.044) British Journal of General Practice, October 2023 e740 possible).This was also supported by the one study to have examined the role of vertebral fractures on self-harm. 23ooled analysis of three studies found that people with osteoporosis were no more likely to die from suicide than patients without osteoporosis.

Strengths and limitations
This is the first paper, to the authors' knowledge, to assemble literature on osteoporosis and fractures as risk factors for self-harm, suicidal ideation, and suicide.Despite only 15 studies being included it was possible to conduct a meta-analysis that provides greater reliability of risk estimates.Another strength is the analysis of fracture as a risk factor by bodily location, which highlighted vertebral fractures as a potentially important risk factor.There are, however, several limitations that need to be recognised.Papers were primarily characterised by population group of either osteoporosis or fractures and, although the poor coding of osteoporosis and fractures in primary care data has previously been noted, 30 the authors undertook quality assurance to assess such variations Table 3.Crude and adjusted risk estimates of fracture and suicide by fracture location Erlangsen et al 13 Erlangsen et al 13 Jang et al 24 Jang et al 28 Jang et al 27 Turvey et al 25  (Supplementary Figure S1).A further limitation of this study is that, because of the variety of definitions, populations, and risk estimates used, it was only possible to pool three studies, which varied widely in quality.However, the fact that the meta-analysis included the study by Chang et al 14 that was a very large study, and, that all estimates were adjusted, adds some credibility to these findings and an opportunity for others to build on.

Comparison with existing literature
Vertebral fractures have significant longterm physical, psychological, and social sequelae.Pain is the likely mechanism driving the association between the presence of a fracture and subsequent self-harm, suicidal ideation, and suicide; especially as osteoporotic fractures have been reported to induce both acute and chronic nociceptive and neuropathic pain. 31,32The presence of fractures has social implications because of the fear of falling or of recurrent fracture, 33,34 with osteoporotic fracture increasing the odds of functional decline by 48%. 35Further poor osteoporotic fracture outcomes include increased social support requirements and diminished quality of life, including depression and deterioration in perceived health. 36,37uicide risk is multifactorial, with depression, anxiety, and other mood disorders as major risk factors.As such, a fracture's impact on patient mental health may in part explain the risk identified related to vertebral fractures. 38Furthermore, the effects of fractures last long beyond the time the injury was sustained.Hallberg et al reported that, 2 years post-fracture, the social function and mental health components of health-related quality of life were still significantly lower than controls in those with vertebral and hip fractures. 39Although this aligns with spinal, and to a lesser extent hip, fractures being significantly associated with suicide in this review, it does not align with the temporal nature of suicide risk post-fracture as reported by Jang et al. 27 They found that risk of suicide was highest in the first 180 days after hip fracture (HR 2.97, 95% CI = 1.32 to 6.69), and then reduced, 24 with similar patterns in pelvic and spinal fractures. 27,28his lack of association with osteoporosis and suicide may relate to the fact that osteoporosis without fracture is primarily asymptomatic and, although low trauma fracture is an indicator of osteoporosis, not all those who have low bone density have broken a bone.As such, osteoporosis is often referred to as 'silent'. 9Despite this common label, a diagnosis of osteoporosis can cause changes in the perception of one's self that may result in social isolation, low mood, and anxiety. 34

Implications for research and practice
This study indicates that patients diagnosed with vertebral fractures might benefit from case finding for mood disorders, such as depression, a risk factor for suicide, and enquiring about suicide and self-harm. 40steoporosis is commonly diagnosed and patients seek help for fracture care in primary care, which may therefore be an important setting to identify patients at risk of self-harm and suicide and intervene early.New research would enable the assessment of the feasibility, acceptability, and effectiveness of such approaches.Early intervention studies have shown that physical exercise programmes that focus on physical anxiety can decrease fear of falls in those with osteoporosis, and potentially lead to lower rates of isolation, 41,42 which might be a causative factor in self-harm and suicide.Such programmes have been shown to be effectively implemented through primary care. 43n conclusion, although studies were typically too diverse for pooling of data, one meta-analysis of three studies showed no association between osteoporosis and suicide.Across several individual studies, vertebral fractures were shown to be potential risk factors for suicide.This review demonstrates the potential importance of teasing apart the role of osteoporosis and fracture in research, and provides strong justification for further research around vertebral fracture in this area.Primary care clinicians could implement case finding for mood disorders and suicide risk assessment in patients with vertebral fractures and undertake subsequent management.

Figure 1 .
Figure 1.PRISMA flow diagram for the study.29

a
No mean and/or SD age of participants provided.b Papers reporting on both osteoporosis and fractures.c Hospital classified as a hospital admission record showing osteoporosis as a primary or contributing diagnosis; community classified as at least two episodes of care recorded for osteoporosis in the physician claim file.d Papers reporting on both self-harm and suicide.GPRD = General Practice Research Database.NHIS = National Health Insurance Service.NRP = National Registry of Patients.SD = standard deviation.VDW = virtual data warehouse.

a
Adjusted odds ratio (Model 1).b Fracture in past 5 years.c Other fracture.d Pathological fracture.e Fracture >50 years.CI = confidence interval.HR = hazard ratio.IR = incidence ratio.

Figure 2 .
Figure 2. Pooled adjusted odds ratio estimates from studies examining the association between osteoporosis and suicide.CI = confidence interval.DL = DerSimonian-Laird.

Table 2 . Reported crude and adjusted risk estimates for osteoporosis and fractures as a risk factor for self-harm, suicidal ideation/behaviour and suicide
et al reported the greatest